[GRASS-dev] [grass-code I][455] r.proj segfaults after Jul 14 changes

[Brad Douglas]

On Fri, 2007-08-10 at 12:07 -0500, William Kyngesburye wrote:
> On Aug 10, 2007, at 11:32 AM, Brad Douglas wrote:
> 
> > On Fri, 2007-08-10 at 08:52 -0500, William Kyngesburye wrote:
> >>
> >> If it is intended to be optional for G_done_msg, then this needs to
> >> be fixed.  It's the vsprintf(buffer,msg,ap); part in there that is
> >> causing trouble on OSX.  If NULL is valid, then a check for NULL
> >> should be done before vsprintf is called.
> >
> > I'm opposed to adding largely unnecessary code.  The console is NOT
> > designed for well laid out display.  It is there for informational
> > purposes only.  IMO, if you want pretty messages, use a GUI.
> >
> It's not a matter of changing the way it prints the message.  The  
> problem is that passing NULL to vsprintf() on OSX causes a bus  
> error.  If NULL is meant to be valid for G_done_msg(), then it needs  
> to trap that so it doesn't pass it to vsprintf().

How many times to I have to state that passing NULL to *printf() va*()
is absolutely, positively **WRONG**?  And we should NOT be modifying the
API to accommodate one person's stylistic choices.

Is anyone listening?

G_msg_done() is not living up to its intended use.  It should probably
be deprecated.


-- 
Brad Douglas <rez touchofmadness com>                    KB8UYR/6
Address: 37.493,-121.924 / WGS84    National Map Corps #TNMC-3785